Loading ...
Sorry, an error occurred while loading the content.

Re: [Xylotex] Microphonics?

Expand Messages
  • Tony Jeffree
    ... This sounds to me to be not entirely unrelated to the noise problems I have been encountering (see Grounding question thread and more detail below).
    Message 1 of 15 , Mar 1, 2004
    • 0 Attachment
      At 03:42 29/02/2004, you wrote:
      >Since my noise problem seems to have been getting progressively worse over
      >the past few weeks of testing, I began to think that maybe there is a
      >problem with the board itself. (The last thing I did last night was to put
      >ferrites on the 26-pin ribbon cable leading from the DB25 to the board, and
      >on the stepper output wires leading from the board to the CB bulkhead
      >connectors--no improvement.)
      >
      >Since I had the chassis top removed, I had the thought to tap components to
      >see if I could induce noise. DB25 connector, nothing. Ribbon cable,
      >nothing. Output wires, nothing. Heatsink on one of the Allegro chips,
      >jerks on the stepper motors. Another heatsink, instant screeching noise
      >from the power supply (going into overload protection, I think) all
      >steppers lost their holding torque and the 24V fan quit. I quickly powered
      >down, let it sit a while and powered back up. Back to normal until I
      >tapped the heatsink, then the repeat of the traumatic condidion.
      >
      >So this afternoon I swapped the board with another board I had also added
      >the filter caps to, powered it up and it also seems to be sensitive to
      >tapping. *But* as long as I don't touch it, there are no random steps on
      >my gantry, even with the spindle motor running. I have yet to run a long
      >test gcode program, but I will do that later tonight and see if the
      >operation is indeed stable over a period of an hour or so.
      >
      >I am wondering if there is a possibility of a shock-sensitive component,
      >marginal cold solder joint or marginally cracked trace on the first board
      >that would explain the seemingly deteriorating condition and sensitivity to
      >tapping (in my experience, PCB assemblies should not be sensitive like
      >that--it reminds me of diagnosing a bad vacuum tube by tapping it--thus the
      >title of this email.)
      >
      >In any case, the first board works well on the Sherline mill--it is only on
      >the Techno gantry that I am seeing the problems, and the second board seems
      >to work well there. So each board for its own machine...

      This sounds to me to be not entirely unrelated to the noise problems I have
      been encountering (see "Grounding question" thread and more detail below).
      Since my last post in that thread, I have completely "fried" my 3-axis
      board (very spectacularly smoked one of the Allegro chips, several of its
      "legs" fused, killed all 3 axes), and the *only* change I had made between
      the "killer" configuration and the previous (working, but noisy)
      configuration was to daisychain the 28V supply from the 3 axis board to the
      1 axis board (it had previously been "star" wired to the 2 boards directly
      from the PSU capacitor). The only conclusion I have come to is that the
      Allegro chip, and/or the board tracks, and/or the ancilliary wiring can
      (and does) pick up any radiated EM energy that is going, and that this can
      result in the false generation of steps by the chip; in the case of my
      "smoked" chip, it may well be that pickup can affect the current sensing
      circuitry too, leading to over-driving the chip and consequent release of
      the "magic smoke". I suspect that the "microphonic" behavior of your
      heatsinks is further evidence of this - if there is a magnetic field for
      the vanes to interact with, vibrating the vanes would change any induced
      current in the heatsink.

      I have been working with two different 4-axis setups, both using a
      combination of DeskCNC and Xylotex cards (the first based on the DeskCNC
      combined controller/3-axis driver card <which uses the same Allegro chips
      as the Xylotex> plus a single axis Xylotex, the second based on the DeskCNC
      controller card plus a 3 axis and a single axis Xylotex), both of which
      exhibited noise problems on one or more axes (although I never "pinged" the
      heatsinks while powered up - my sense of self-preservation prevents me from
      putting fingers inside a live box). I have eventually concluded that these
      Allegro-based controllers don't like to be mounted in close proximity to
      any potential noise or EM energy source, even ones that you would normally
      regard as not a problem, such as the PSU transformer. With one
      configuration variant I tested, the 4th axis would "free run" the motor,
      perfectly smoothly, at what appeared to be a step rate of 50 Hz (UK mains
      frequency)!

      After "smoking" the 3-axis Xylotex (as above), I re-assembled the first
      (3-axis DeskCNC + 1 axis Xylotex) configuration, but instead of using the
      home-built PSU that had ben in the same box as the cards, I used a
      metal-cased bench PSU (25V, 2.5A) that was a couple of feet away from the
      boards, and the boards were simply laid on the (non-conducting!) bench. All
      4 axes behaved perfectly - no random steps, no "free running" axes). I have
      yet to test this with the 3+1 Xylotex setup, as I am now short of a 3-axis
      board and am reluctant to spend yet more money chasing this problem, and
      the UK distributor is understandably reluctant to replace my fried board.

      Can you tell me:

      a) How close to the PSU transformer has your board been mounted?
      b) What type of transformer (torroidal, or conventional "square" frame) you
      are using? (the relevance being that torroidal transformers generally seem
      to radiate less EM energy than conventional transformers.)
      c) Have you tried physically separating the PSU from the board (by a couple
      of feet) & seeing if that makes any difference? (has anyone else done this?)

      I must confess that I am suffering major sense of humour failure over this
      whole issue - I have now destroyed a total of four 3-axis Xylotex boards in
      my quest to construct a working controller, and in only one of those cases
      was it due to an obvious mistake on my part (inadvertently connecting the
      high voltage supply to the 5V pins of the board - NOT, I hasten to add, a
      mistake that I have repeated!). Something of a pity as I was in the process
      of writing up my experiences with this board and with DeskCNC for an
      article for Model Engineers' Workshop magazine. I guess I may still do so,
      but in the form of a "cautionary tale".


      Regards,
      Tony
    • broken003@juno.com
      ... I must confess that I am suffering major sense of humour failure over this whole issue - I have now destroyed a total of four 3-axis Xylotex boards in my
      Message 2 of 15 , Mar 1, 2004
      • 0 Attachment
        ---Snip---
        "I must confess that I am suffering major sense of humour failure over this
        whole issue - I have now destroyed a total of four 3-axis Xylotex boards in
        my quest to construct a working controller, and in only one of those cases
        was it due to an obvious mistake on my part"
        -----Snip---


        Can you share the other methods that caused the board to fail? If you did already can you send a link to the poist? I believe that I am beginning the process of diagnosing "odd" behaviour from my board and I dont want to make the same mistakes you did.


        Thanks
        Mark









        -- Tony Jeffree <tony@...> wrote:
        At 03:42 29/02/2004, you wrote:
        >Since my noise problem seems to have been getting progressively worse over
        >the past few weeks of testing, I began to think that maybe there is a
        >problem with the board itself. (The last thing I did last night was to put
        >ferrites on the 26-pin ribbon cable leading from the DB25 to the board, and
        >on the stepper output wires leading from the board to the CB bulkhead
        >connectors--no improvement.)
        >
        >Since I had the chassis top removed, I had the thought to tap components to
        >see if I could induce noise. DB25 connector, nothing. Ribbon cable,
        >nothing. Output wires, nothing. Heatsink on one of the Allegro chips,
        >jerks on the stepper motors. Another heatsink, instant screeching noise
        >from the power supply (going into overload protection, I think) all
        >steppers lost their holding torque and the 24V fan quit. I quickly powered
        >down, let it sit a while and powered back up. Back to normal until I
        >tapped the heatsink, then the repeat of the traumatic condidion.
        >
        >So this afternoon I swapped the board with another board I had also added
        >the filter caps to, powered it up and it also seems to be sensitive to
        >tapping. *But* as long as I don't touch it, there are no random steps on
        >my gantry, even with the spindle motor running. I have yet to run a long
        >test gcode program, but I will do that later tonight and see if the
        >operation is indeed stable over a period of an hour or so.
        >
        >I am wondering if there is a possibility of a shock-sensitive component,
        >marginal cold solder joint or marginally cracked trace on the first board
        >that would explain the seemingly deteriorating condition and sensitivity to
        >tapping (in my experience, PCB assemblies should not be sensitive like
        >that--it reminds me of diagnosing a bad vacuum tube by tapping it--thus the
        >title of this email.)
        >
        >In any case, the first board works well on the Sherline mill--it is only on
        >the Techno gantry that I am seeing the problems, and the second board seems
        >to work well there. So each board for its own machine...

        This sounds to me to be not entirely unrelated to the noise problems I have
        been encountering (see "Grounding question" thread and more detail below).
        Since my last post in that thread, I have completely "fried" my 3-axis
        board (very spectacularly smoked one of the Allegro chips, several of its
        "legs" fused, killed all 3 axes), and the *only* change I had made between
        the "killer" configuration and the previous (working, but noisy)
        configuration was to daisychain the 28V supply from the 3 axis board to the
        1 axis board (it had previously been "star" wired to the 2 boards directly
        from the PSU capacitor). The only conclusion I have come to is that the
        Allegro chip, and/or the board tracks, and/or the ancilliary wiring can
        (and does) pick up any radiated EM energy that is going, and that this can
        result in the false generation of steps by the chip; in the case of my
        "smoked" chip, it may well be that pickup can affect the current sensing
        circuitry too, leading to over-driving the chip and consequent release of
        the "magic smoke". I suspect that the "microphonic" behavior of your
        heatsinks is further evidence of this - if there is a magnetic field for
        the vanes to interact with, vibrating the vanes would change any induced
        current in the heatsink.

        I have been working with two different 4-axis setups, both using a
        combination of DeskCNC and Xylotex cards (the first based on the DeskCNC
        combined controller/3-axis driver card <which uses the same Allegro chips
        as the Xylotex> plus a single axis Xylotex, the second based on the DeskCNC
        controller card plus a 3 axis and a single axis Xylotex), both of which
        exhibited noise problems on one or more axes (although I never "pinged" the
        heatsinks while powered up - my sense of self-preservation prevents me from
        putting fingers inside a live box). I have eventually concluded that these
        Allegro-based controllers don't like to be mounted in close proximity to
        any potential noise or EM energy source, even ones that you would normally
        regard as not a problem, such as the PSU transformer. With one
        configuration variant I tested, the 4th axis would "free run" the motor,
        perfectly smoothly, at what appeared to be a step rate of 50 Hz (UK mains
        frequency)!

        After "smoking" the 3-axis Xylotex (as above), I re-assembled the first
        (3-axis DeskCNC + 1 axis Xylotex) configuration, but instead of using the
        home-built PSU that had ben in the same box as the cards, I used a
        metal-cased bench PSU (25V, 2.5A) that was a couple of feet away from the
        boards, and the boards were simply laid on the (non-conducting!) bench. All
        4 axes behaved perfectly - no random steps, no "free running" axes). I have
        yet to test this with the 3+1 Xylotex setup, as I am now short of a 3-axis
        board and am reluctant to spend yet more money chasing this problem, and
        the UK distributor is understandably reluctant to replace my fried board.

        Can you tell me:

        a) How close to the PSU transformer has your board been mounted?
        b) What type of transformer (torroidal, or conventional "square" frame) you
        are using? (the relevance being that torroidal transformers generally seem
        to radiate less EM energy than conventional transformers.)
        c) Have you tried physically separating the PSU from the board (by a couple
        of feet) & seeing if that makes any difference? (has anyone else done this?)

        I must confess that I am suffering major sense of humour failure over this
        whole issue - I have now destroyed a total of four 3-axis Xylotex boards in
        my quest to construct a working controller, and in only one of those cases
        was it due to an obvious mistake on my part (inadvertently connecting the
        high voltage supply to the 5V pins of the board - NOT, I hasten to add, a
        mistake that I have repeated!). Something of a pity as I was in the process
        of writing up my experiences with this board and with DeskCNC for an
        article for Model Engineers' Workshop magazine. I guess I may still do so,
        but in the form of a "cautionary tale".


        Regards,
        Tony





        Yahoo! Groups Links






        ________________________________________________________________
        The best thing to hit the Internet in years - Juno SpeedBand!
        Surf the Web up to FIVE TIMES FASTER!
        Only $14.95/ month - visit www.juno.com to sign up today!
      • Tony Jeffree
        ... I can understand that! In the other two cases, I have no real idea to this day what the source of the failure was, although the setup there may shed more
        Message 3 of 15 , Mar 1, 2004
        • 0 Attachment
          At 11:48 01/03/2004, you wrote:
          >Can you share the other methods that caused the board to fail? If you did
          >already can you send a link to the poist? I believe that I am beginning
          >the process of diagnosing "odd" behaviour from my board and I dont want to
          >make the same mistakes you did.

          I can understand that!

          In the other two cases, I have no real idea to this day what the source of
          the failure was, although the setup there may shed more light on the noise
          problems (see below). My earliest attempts to get the board running
          involved two separate power supplies (two transformers...etc), one
          supplying the motor drive voltage (same bench PSU as in the more recent
          tests), and the second supplying the logic 5V via a 7805 regulator, and
          just the 3-axis Xylotex board; both PSUs were remote from the board (again,
          about 2-3 feet of power cable). I got the board running, controlling my
          Taig mill; all seemed to be fine until one axis simply stopped working. I
          shipped it back to Jeff for repair; it then worked again for long enough to
          perform some fairly lengthy tasks (for example, milling a couple of small
          PCBs); the next day when I powered the board up, only one axis was working,
          the other two wouldn't drive the motors at all. The only peculiarity about
          this setup was the fact that the two PSUs were powered up independently of
          each other; I understand from talking to the distributor in the UK that the
          boards shouldn't be sensitive to the order in which the supply rails power
          up, however I did notice that if you apply the motor voltage first, there
          is an initial power surge visible on the bench PSU.

          However, in both cases, the setup didn't exhibit the "random noise"
          syndrome, except for the fact that turning on/off fluorescent light
          fittings in the workshop caused a few random steps on one or more axes. Of
          course, this isn't comparing like with like, as the later tests were with 4
          axes, not 3.


          Regards,
          Tony
        • xylotex
          Hi Tony, There hasn t been a fried board that I haven t been able to fix yet. E-mail me. Jeff ... problems I have ... below). ... axis ... of its ...
          Message 4 of 15 , Mar 1, 2004
          • 0 Attachment
            Hi Tony,

            There hasn't been a 'fried' board that I haven't been able to fix
            yet. E-mail me.

            Jeff

            >
            > This sounds to me to be not entirely unrelated to the noise
            problems I have
            > been encountering (see "Grounding question" thread and more detail
            below).
            > Since my last post in that thread, I have completely "fried" my 3-
            axis
            > board (very spectacularly smoked one of the Allegro chips, several
            of its
            > "legs" fused, killed all 3 axes), and the *only* change I had made
            between
            > the "killer" configuration and the previous (working, but noisy)
            > configuration was to daisychain the 28V supply from the 3 axis
            board to the
            > 1 axis board (it had previously been "star" wired to the 2 boards
            directly
            > from the PSU capacitor). The only conclusion I have come to is that
            the
            > Allegro chip, and/or the board tracks, and/or the ancilliary wiring
            can
            > (and does) pick up any radiated EM energy that is going, and that
            this can
            > result in the false generation of steps by the chip; in the case of
            my
            > "smoked" chip, it may well be that pickup can affect the current
            sensing
            > circuitry too, leading to over-driving the chip and consequent
            release of
            > the "magic smoke". I suspect that the "microphonic" behavior of
            your
            > heatsinks is further evidence of this - if there is a magnetic
            field for
            > the vanes to interact with, vibrating the vanes would change any
            induced
            > current in the heatsink.
            >
            > I have been working with two different 4-axis setups, both using a
            > combination of DeskCNC and Xylotex cards (the first based on the
            DeskCNC
            > combined controller/3-axis driver card <which uses the same Allegro
            chips
            > as the Xylotex> plus a single axis Xylotex, the second based on the
            DeskCNC
            > controller card plus a 3 axis and a single axis Xylotex), both of
            which
            > exhibited noise problems on one or more axes (although I
            never "pinged" the
            > heatsinks while powered up - my sense of self-preservation prevents
            me from
            > putting fingers inside a live box). I have eventually concluded
            that these
            > Allegro-based controllers don't like to be mounted in close
            proximity to
            > any potential noise or EM energy source, even ones that you would
            normally
            > regard as not a problem, such as the PSU transformer. With one
            > configuration variant I tested, the 4th axis would "free run" the
            motor,
            > perfectly smoothly, at what appeared to be a step rate of 50 Hz (UK
            mains
            > frequency)!
            >
            > After "smoking" the 3-axis Xylotex (as above), I re-assembled the
            first
            > (3-axis DeskCNC + 1 axis Xylotex) configuration, but instead of
            using the
            > home-built PSU that had ben in the same box as the cards, I used a
            > metal-cased bench PSU (25V, 2.5A) that was a couple of feet away
            from the
            > boards, and the boards were simply laid on the (non-conducting!)
            bench. All
            > 4 axes behaved perfectly - no random steps, no "free running"
            axes). I have
            > yet to test this with the 3+1 Xylotex setup, as I am now short of a
            3-axis
            > board and am reluctant to spend yet more money chasing this
            problem, and
            > the UK distributor is understandably reluctant to replace my fried
            board.
            >
            > Can you tell me:
            >
            > a) How close to the PSU transformer has your board been mounted?
            > b) What type of transformer (torroidal, or conventional "square"
            frame) you
            > are using? (the relevance being that torroidal transformers
            generally seem
            > to radiate less EM energy than conventional transformers.)
            > c) Have you tried physically separating the PSU from the board (by
            a couple
            > of feet) & seeing if that makes any difference? (has anyone else
            done this?)
            >
            > I must confess that I am suffering major sense of humour failure
            over this
            > whole issue - I have now destroyed a total of four 3-axis Xylotex
            boards in
            > my quest to construct a working controller, and in only one of
            those cases
            > was it due to an obvious mistake on my part (inadvertently
            connecting the
            > high voltage supply to the 5V pins of the board - NOT, I hasten to
            add, a
            > mistake that I have repeated!). Something of a pity as I was in the
            process
            > of writing up my experiences with this board and with DeskCNC for
            an
            > article for Model Engineers' Workshop magazine. I guess I may still
            do so,
            > but in the form of a "cautionary tale".
            >
            >
            > Regards,
            > Tony
          • xylotex
            Hi Tony, I read in Allegro literature somewhere - I don t know which document - that the order the power supplies are turned on (Vbb & Vcc) should not make a
            Message 5 of 15 , Mar 1, 2004
            • 0 Attachment
              Hi Tony,
              I read in Allegro literature somewhere - I don't know which
              document - that the order the power supplies are turned on (Vbb &
              Vcc) should not make a difference. For the high side gates to be
              active, the internal oscillators need to be running. This should
              only be happening with power to the Vcc. With that said, I always
              recommend turning on the Vcc first (or at the same time as Vbb).
              This allows internal clocks, counters, oscillators, etc. to become
              properly setup before Vbb is active. In a supply that has Vcc
              generated from Vbb (like 7805), as the Vbb starts to rise past the
              threshold of the 7085 to make a stable 5V (about 7 or 8 volts) the
              internal circuitry of the Allegro should be active and stable. The
              onboard reset circuit (on the single & three axes boards) will hold
              the device (Allegro) in a reset state for 140 mS after the Vcc
              voltage rises past 4.63V. Since the chip can actually run with
              voltages as low as 3.3V, internal circuitry should be more than
              stable by the time the reset circuit disengages.
              There are two basic ways to destroy a chip. Overvoltage and
              overcurrent. It seems apparent that the 35V absolute maximum
              published by Allegro is just that: absolute maximum. Bad things
              happen when you start to go past this. The suggested running maximum
              is about 15% less than the abs. max., or about 30VDC. My guess is
              that you are not having an overvoltage problem. The other
              destructive force mentioned is current. If the overcurrent happens
              slowly, the chip will heat up, and the internal overheat protection
              will shut it down before it destroys itself. When the temperature
              has dropped, it will turn on again. Consistant overcurrent/overheat -
              shutdown cycle can dastrically reduce the life of the chip. Again,
              probably not what you are experiencing. Finally there is a
              more "instantaneous" current surge. These would be caused by not
              shutting down the PWM gate when it is supposed to be shut down. The
              gate should turn off after the internal circuitry senses that the
              current has passed its' threshold (by measuring the voltage across
              the sense resistor). An RC circuit, part of a "one-shot" circuit is
              used to keep the driver off for a fixed amount of time once the sense
              circuit says the proper current level has been reached. These two
              things are the most likely places for overcurrent error. If the
              sense circuit does not sense properly, it may "think" the current
              never reaches full, thus always stays on. If the RC circuit fails to
              trigger the one-shot, and turn off the drive, current remains on.
              It would appear from your previous posts, that noise is being
              injected into one of the two above mentioned circuits. This combined
              with Randy's "microphonics" information leads me to believe that it
              would be the RC circuit that is causing the problem (rather than the
              sense circuit). These circuits (RC/one-shot) are also responsible
              for most of the noise you hear from the motor (I liken it to water
              running through a pipe, others hear growling and groaning). The
              problem now becomes identifying how/where the noise from your power
              supply is being injected into the RC/one-shot (if this actaully is
              the case). Since you mentioned that the supply can actually make the
              drive step at about 50Hz (your mains), it would appear as though your
              supply it sending out enough energy to falsely trigger the STEP
              line. Although the STEP line has a large antenna (the 6 to 8 inch
              long IDC26-DB25 cable), enough noise seems to be being generated to
              disrupt the RC/one-shot circuit as well (two apparently separate
              issues that may come from the same place). Perhaps this same noise
              is being injected in to the motor cables. Eventually the cables lead
              to onboard traces that end up near the chips' RC/one-shot circuit
              (well everthing eventually does as the chip is only about half inch
              square). Without reviewing previous post, my guess is that you have
              sheilded & grounded motor cables, whcih should prevent alot of that
              from happening anyway. Barring that, your suggestion of distance
              (moving the supply away from the drive) seems to be the only viable
              (although not generally practical) solution.

              Does this info shed any light on to what you are seeing?

              Thanks,

              Jeff

              >
              > I can understand that!
              >
              > In the other two cases, I have no real idea to this day what the
              source of
              > the failure was, although the setup there may shed more light on
              the noise
              > problems (see below). My earliest attempts to get the board running
              > involved two separate power supplies (two transformers...etc), one
              > supplying the motor drive voltage (same bench PSU as in the more
              recent
              > tests), and the second supplying the logic 5V via a 7805
              regulator, and
              > just the 3-axis Xylotex board; both PSUs were remote from the board
              (again,
              > about 2-3 feet of power cable). I got the board running,
              controlling my
              > Taig mill; all seemed to be fine until one axis simply stopped
              working. I
              > shipped it back to Jeff for repair; it then worked again for long
              enough to
              > perform some fairly lengthy tasks (for example, milling a couple of
              small
              > PCBs); the next day when I powered the board up, only one axis was
              working,
              > the other two wouldn't drive the motors at all. The only
              peculiarity about
              > this setup was the fact that the two PSUs were powered up
              independently of
              > each other; I understand from talking to the distributor in the UK
              that the
              > boards shouldn't be sensitive to the order in which the supply
              rails power
              > up, however I did notice that if you apply the motor voltage first,
              there
              > is an initial power surge visible on the bench PSU.
              >
              > However, in both cases, the setup didn't exhibit the "random noise"
              > syndrome, except for the fact that turning on/off fluorescent light
              > fittings in the workshop caused a few random steps on one or more
              axes. Of
              > course, this isn't comparing like with like, as the later tests
              were with 4
              > axes, not 3.
              >
              >
              > Regards,
              > Tony
            • broken003@juno.com
              Can Jeff or someone else summarize the optimal method to mount/connect the xylotex and related components. I plan to change my layout to hopefully eradicate
              Message 6 of 15 , Mar 1, 2004
              • 0 Attachment
                Can Jeff or someone else summarize the optimal method to mount/connect the xylotex and related components. I plan to change my layout to hopefully eradicate the demons that have been infecting my system.

                From what I read, some suggestions are (correct me if I am wrong)
                keep xylotex board away from everything
                mounted on non conductive surface
                keep anything that generates a rf signal on a different circuit
                Ferrite cores aren't a bad idea if needed
                Stepper wires grounded on one end,


                any other suggestions?


                Thanks
                Mark

                Please note: message attached



                ________________________________________________________________
                The best thing to hit the Internet in years - Juno SpeedBand!
                Surf the Web up to FIVE TIMES FASTER!
                Only $14.95/ month - visit www.juno.com to sign up today!
              • Tony Jeffree
                ... OK - so what this amounts to is that using a logic supply derived from the high side supply (via a 7805 or similar) is a smart move, although this may not
                Message 7 of 15 , Mar 1, 2004
                • 0 Attachment
                  At 17:40 01/03/2004, you wrote:
                  >Hi Tony,
                  > I read in Allegro literature somewhere - I don't know which
                  >document - that the order the power supplies are turned on (Vbb &
                  >Vcc) should not make a difference. For the high side gates to be
                  >active, the internal oscillators need to be running. This should
                  >only be happening with power to the Vcc. With that said, I always
                  >recommend turning on the Vcc first (or at the same time as Vbb).
                  >This allows internal clocks, counters, oscillators, etc. to become
                  >properly setup before Vbb is active. In a supply that has Vcc
                  >generated from Vbb (like 7805), as the Vbb starts to rise past the
                  >threshold of the 7085 to make a stable 5V (about 7 or 8 volts) the
                  >internal circuitry of the Allegro should be active and stable. The
                  >onboard reset circuit (on the single & three axes boards) will hold
                  >the device (Allegro) in a reset state for 140 mS after the Vcc
                  >voltage rises past 4.63V. Since the chip can actually run with
                  >voltages as low as 3.3V, internal circuitry should be more than
                  >stable by the time the reset circuit disengages.

                  OK - so what this amounts to is that using a logic supply derived from the
                  high side supply (via a 7805 or similar) is a smart move, although this may
                  not be essential.

                  > There are two basic ways to destroy a chip. Overvoltage and
                  >overcurrent. It seems apparent that the 35V absolute maximum
                  >published by Allegro is just that: absolute maximum. Bad things
                  >happen when you start to go past this. The suggested running maximum
                  >is about 15% less than the abs. max., or about 30VDC. My guess is
                  >that you are not having an overvoltage problem.

                  I would agree with that - the PSU I built is around 27V and has a 33V Zener
                  across the cap, so over-voltage is very unlikely to be the issue here. The
                  bench PSU that I have also used is variable voltage up to 25V & regulated,
                  so again, that would not have created an over voltage.

                  > The other
                  >destructive force mentioned is current. If the overcurrent happens
                  >slowly, the chip will heat up, and the internal overheat protection
                  >will shut it down before it destroys itself. When the temperature
                  >has dropped, it will turn on again. Consistant overcurrent/overheat -
                  > shutdown cycle can dastrically reduce the life of the chip. Again,
                  >probably not what you are experiencing.

                  Agreed. It fried instantaneously on power up.

                  >Finally there is a
                  >more "instantaneous" current surge. These would be caused by not
                  >shutting down the PWM gate when it is supposed to be shut down. The
                  >gate should turn off after the internal circuitry senses that the
                  >current has passed its' threshold (by measuring the voltage across
                  >the sense resistor). An RC circuit, part of a "one-shot" circuit is
                  >used to keep the driver off for a fixed amount of time once the sense
                  >circuit says the proper current level has been reached. These two
                  >things are the most likely places for overcurrent error. If the
                  >sense circuit does not sense properly, it may "think" the current
                  >never reaches full, thus always stays on. If the RC circuit fails to
                  >trigger the one-shot, and turn off the drive, current remains on.
                  > It would appear from your previous posts, that noise is being
                  >injected into one of the two above mentioned circuits.

                  That would be consistent with what I saw happening.

                  >This combined
                  >with Randy's "microphonics" information leads me to believe that it
                  >would be the RC circuit that is causing the problem (rather than the
                  >sense circuit). These circuits (RC/one-shot) are also responsible
                  >for most of the noise you hear from the motor (I liken it to water
                  >running through a pipe, others hear growling and groaning). The
                  >problem now becomes identifying how/where the noise from your power
                  >supply is being injected into the RC/one-shot (if this actaully is
                  >the case). Since you mentioned that the supply can actually make the
                  >drive step at about 50Hz (your mains), it would appear as though your
                  >supply it sending out enough energy to falsely trigger the STEP
                  >line. Although the STEP line has a large antenna (the 6 to 8 inch
                  >long IDC26-DB25 cable), enough noise seems to be being generated to
                  >disrupt the RC/one-shot circuit as well (two apparently separate
                  >issues that may come from the same place).

                  I 'scoped the step line while the drive was generating random pulses, and
                  it was absolutely dead flat. So my conclusion from this is that the point
                  of noise injection was not the step/direction input lines themselves, but
                  somewhere else on the board or within the chip.

                  > Perhaps this same noise
                  >is being injected in to the motor cables. Eventually the cables lead
                  >to onboard traces that end up near the chips' RC/one-shot circuit
                  >(well everthing eventually does as the chip is only about half inch
                  >square). Without reviewing previous post, my guess is that you have
                  >sheilded & grounded motor cables, whcih should prevent alot of that
                  >from happening anyway.

                  I am using shielded motor cables, also shielded signal cables for the
                  step/direction lines (with the shields of the latter grounded to logic 0V).

                  >Barring that, your suggestion of distance
                  >(moving the supply away from the drive) seems to be the only viable
                  >(although not generally practical) solution.

                  I believe that is probably the case, and is what I will try next. A more
                  generally practical solution might be to isolate the PSU within its own
                  shielded compartment within a larger enclosure - basically stick it inside
                  its own Faraday cage - so that the whole of the electronics can be kept in
                  one box without the PSU affecting the controller. Again, this might be
                  worth some experimentation.


                  > Does this info shed any light on to what you are seeing?

                  Very illuminating - thanks - and largely confirms to me that what I suspect
                  is happening is at least possible (although whether I am right or not
                  remains to be seen!)

                  There's a potentially interesting experiment to be done here, which I am
                  reluctant to do myself owing to the possibility of destroying another
                  board, but it would be fascinating to see what the effect is of energising
                  a coil with 50 Hz (or 60 Hz) and seeing what happens as you bring it
                  towards a working controller (powered up, connected to motors, but not
                  being stepped). If the result is that the controller starts to generate
                  random steps, or worse, smokes one of the chips,then maybe we have the culprit.

                  Regards,
                  Tony


                  >Thanks,
                  >
                  >Jeff
                  >
                  > >
                  > > I can understand that!
                  > >
                  > > In the other two cases, I have no real idea to this day what the
                  >source of
                  > > the failure was, although the setup there may shed more light on
                  >the noise
                  > > problems (see below). My earliest attempts to get the board running
                  > > involved two separate power supplies (two transformers...etc), one
                  > > supplying the motor drive voltage (same bench PSU as in the more
                  >recent
                  > > tests), and the second supplying the logic 5V via a 7805
                  >regulator, and
                  > > just the 3-axis Xylotex board; both PSUs were remote from the board
                  >(again,
                  > > about 2-3 feet of power cable). I got the board running,
                  >controlling my
                  > > Taig mill; all seemed to be fine until one axis simply stopped
                  >working. I
                  > > shipped it back to Jeff for repair; it then worked again for long
                  >enough to
                  > > perform some fairly lengthy tasks (for example, milling a couple of
                  >small
                  > > PCBs); the next day when I powered the board up, only one axis was
                  >working,
                  > > the other two wouldn't drive the motors at all. The only
                  >peculiarity about
                  > > this setup was the fact that the two PSUs were powered up
                  >independently of
                  > > each other; I understand from talking to the distributor in the UK
                  >that the
                  > > boards shouldn't be sensitive to the order in which the supply
                  >rails power
                  > > up, however I did notice that if you apply the motor voltage first,
                  >there
                  > > is an initial power surge visible on the bench PSU.
                  > >
                  > > However, in both cases, the setup didn't exhibit the "random noise"
                  > > syndrome, except for the fact that turning on/off fluorescent light
                  > > fittings in the workshop caused a few random steps on one or more
                  >axes. Of
                  > > course, this isn't comparing like with like, as the later tests
                  >were with 4
                  > > axes, not 3.
                  > >
                  > >
                  > > Regards,
                  > > Tony
                  >
                  >
                  >
                  >
                  >
                  >Yahoo! Groups Links
                  >
                  >
                  >
                  >

                  Regards,
                  Tony
                • broken003@juno.com
                  Can Jeff or someone else summarize the optimal method to mount/connect the xylotex and related components. I plan to change my layout to hopefully eradicate
                  Message 8 of 15 , Mar 1, 2004
                  • 0 Attachment
                    Can Jeff or someone else summarize the optimal method to mount/connect the xylotex and related components. I plan to change my layout to hopefully eradicate the demons that have been infecting my system.

                    From what I read, some suggestions are (correct me if I am wrong)
                    keep xylotex board away from everything
                    mounted on non conductive surface
                    keep anything that generates a rf signal on a different circuit
                    Ferrite cores aren't a bad idea if needed
                    Stepper wires grounded on one end,


                    any other suggestions?


                    Thanks
                    Mark

                    Please note: message attached



                    ________________________________________________________________
                    The best thing to hit the Internet in years - Juno SpeedBand!
                    Surf the Web up to FIVE TIMES FASTER!
                    Only $14.95/ month - visit www.juno.com to sign up today!
                  • Fred Walter
                    ... I second this request. With photos, if at all possible. I haven t attached stepper motors to anything yet, but I ve tried out a bunch of them, to make sure
                    Message 9 of 15 , Mar 2, 2004
                    • 0 Attachment
                      broken003@j... wrote:
                      > Can Jeff or someone else summarize the optimal method
                      > to mount/connect the xylotex and related components.

                      I second this request. With photos, if at all possible.

                      I haven't attached stepper motors to anything yet,
                      but I've tried out a bunch of them, to make sure that they work.
                      With my setup (see the photos section) stepper motors attached
                      to the the X-axis occassionally don't turn smoothly - they 'jump'.
                      If you are holding the motor in your hand, there is a bit of extra
                      torque for a split second. I'm assuming this is a missed step. :-(

                      I haven't gotten around to fixing the problem,
                      because I need to finish other things off first.
                    • broken003
                      Somebody? Anybody? I would really like to get this discussion going. I started diagnosing my problem last night and my X axis was more likely to stall or
                      Message 10 of 15 , Mar 4, 2004
                      • 0 Attachment
                        Somebody? Anybody? I would really like to get this discussion going.

                        I started diagnosing my problem last night and my X axis was more
                        likely to stall or miss steps with less items hooked to the power
                        supply.

                        for example, unhooking the 4th axis caused X to stall easier



                        --- In Xylotex@yahoogroups.com, "Fred Walter" <canadian_fred@y...>
                        wrote:
                        > broken003@j... wrote:
                        > > Can Jeff or someone else summarize the optimal method
                        > > to mount/connect the xylotex and related components.
                        >
                        > I second this request. With photos, if at all possible.
                        >
                        > I haven't attached stepper motors to anything yet,
                        > but I've tried out a bunch of them, to make sure that they work.
                        > With my setup (see the photos section) stepper motors attached
                        > to the the X-axis occassionally don't turn smoothly - they 'jump'.
                        > If you are holding the motor in your hand, there is a bit of extra
                        > torque for a split second. I'm assuming this is a missed step. :-(
                        >
                        > I haven't gotten around to fixing the problem,
                        > because I need to finish other things off first.
                      • xylotex
                        Hi Mark, Well, it s hard to suggest things for you to do without knowing anything about your current setup. Since everyone will be using a wide variety or
                        Message 11 of 15 , Mar 4, 2004
                        • 0 Attachment
                          Hi Mark,
                          Well, it's hard to suggest things for you to do without knowing
                          anything about your current setup.
                          Since everyone will be using a wide variety or stepper motors,
                          power supplies, enclosures, CPU systems, and mechanical assemblies, a
                          solution for one system will not necessarily work for another.
                          Point in case is Tony's system with the power supply next to the
                          board for his Taig: problems. My system with a switching power
                          supply generating the 24VDC & 5VDC for my Sherline: no problems.
                          If you have a specific setup you would like to discuss, by all
                          means post the details. Please include some specifics like your PC
                          type & speed, the software being used. Your type and rating of
                          stepper motor, how long the stepper cables are, etc. What you are
                          driving (Sherline, Taig, homebrew...), and leadscrew pitch/type.
                          Your power supply rating (both voltage and current). Any sheilding
                          and enclosure you currently have. The length and type of parallel
                          port extension cable. How things are grounded. The Vref setting you
                          are using. The step mode you are using (1/8th step??). Your program
                          parameters like accel and top speed. If you are using a fan.

                          We can go from there.

                          Jeff

                          --- In Xylotex@yahoogroups.com, "broken003" <broken003@j...> wrote:
                          > Somebody? Anybody? I would really like to get this discussion
                          going.
                          >
                          > I started diagnosing my problem last night and my X axis was more
                          > likely to stall or miss steps with less items hooked to the power
                          > supply.
                          >
                          > for example, unhooking the 4th axis caused X to stall easier
                          >
                          >
                          >
                          > --- In Xylotex@yahoogroups.com, "Fred Walter" <canadian_fred@y...>
                          > wrote:
                          > > broken003@j... wrote:
                          > > > Can Jeff or someone else summarize the optimal method
                          > > > to mount/connect the xylotex and related components.
                          > >
                          > > I second this request. With photos, if at all possible.
                          > >
                          > > I haven't attached stepper motors to anything yet,
                          > > but I've tried out a bunch of them, to make sure that they work.
                          > > With my setup (see the photos section) stepper motors attached
                          > > to the the X-axis occassionally don't turn smoothly - they 'jump'.
                          > > If you are holding the motor in your hand, there is a bit of extra
                          > > torque for a split second. I'm assuming this is a missed step. :-(
                          > >
                          > > I haven't gotten around to fixing the problem,
                          > > because I need to finish other things off first.
                        Your message has been successfully submitted and would be delivered to recipients shortly.